Active Control of Boundary-Layer Transition in Laminar Separation Bubbles

Active flow control of laminar separation bubbles (LSBs) is explored in a combined approach based on high quality wind-tunnel experiments and high-fidelity direct numerical simulations (DNS). The favorable to adverse pressure gradient under an inverted modified NACA 64₃ − 618 airfoil generates a separation bubble on the flat plate. Periodic disturbances from an AC-DBD plasma actuator in the experiment and wall normal velocity perturbations in the DNS enhance the development of the dominant shear layer mode in the LSB. The resulting coherent 2D roller structures can delay transition and facilitate mixing in the LSB leading to earlier reattachment. At tested amplitudes, forcing upstream of the onset of the adverse pressure gradient (x_f = 18.75) does not prevent immediate breakdown to turbulence inside the LSB. Increased forcing amplitude was realized by moving actuation closer to the onset of adverse pressure gradient (APG) (x_f = 19.75). Secondary instability analysis (SIA) and DNS without free-stream turbulence (FST) predict significant delay of transition far downstream for moderate forcing amplitudes just below A_cr = 2.5% of the local free-stream velocity. In the presence of FST the 2D mode is modulated with a secondary oblique mode and transition is observed just downstream of mean reattachment in the DNS. The experiment shows modulation of the initial 2D shear layer instability by a 3D mode of the same wavelength. Similar dynamics are observed in the spatial wavelength (= 1) and temporal frequency (f = f_AFC /2 = 100Hz). However, slight differences in the mean flow downstream of reattachment and power spectral density, suggests earlier transition in the experiment compared to DNS results with FST. Variation of the forcing amplitude in the experiment show changes in dominant modes inside the LSB. At higher forcing amplitudes the most dominant mode is no longer two-dimensional, and the second mode shows significant reduction in spanwise wavelength. This suggests the existence of an optimal forcing amplitude in the experiment for the actuator under investigation.